目的 基于体表心电图指标及临床指标构建心房颤动伴心力衰竭患者射频消融（radiofrequency catheter ablation， RFCA）治疗后复发的列线图模型并进行验证。 方法 选取2019年1月至2024年1月期间新乡市中心医院收治的305例房颤伴心力衰竭患者，按照7∶3比例随机分为训练集（213例）和验证集（92例），所有患者均接受RFCA治疗并完成至少1年的随访，根据随访结果分为复发组和未复发组，收集两组体表心电图指标及临床指标。采用多因素logistic回归分析筛选RFCA治疗后复发的危险因素，基于危险因素构建风险预测列线图模型，并对构建的模型进行评估。 结果 共收集305例患者的临床资料，治疗后复发84例（27.54%）；按照7∶3比例随机分为训练集（213例）和验证集（92例），其中训练集复发61例，未复发152例。训练集与验证集临床资料比较差异均无统计学意义（P > 0.05）。训练集复发组持续性房颤占比更大，且复发组CHA2DS2-VASc评分、左心房内径、PR间期、NLR、NT-proBNP水平均高于未复发组（P < 0.05）。多因素逐步回归分析显示，高CHA2DS2-VASc评分、左心房内径大、PR间期延长、高NLR是RFCA治疗后复发的独立危险因素（P < 0.05）。建立4因子预测模型Ln（P/1-P） = -12.87 + 0.84*CHA2DS2-VASc评分 + 0.11*左心房内径 + 0.03*PR间期 + 0.31*NLR，训练集和验证集模型的曲线下面积（AUC）分别为0.85（95% CI： 0.80 ~ 0.91）、0.85（95% CI： 0.76 ~ 0.94）提示该模型预测效能良好；对模型进行Hosmer-Lemeshow拟合优度检验，认为拟合的概率值和实际的概率值基本一致（χ² = 2.43、5.30，P = 0.965、0.725），进一步通过Bootstrap重复抽样1 000次后绘制校准曲线，发现训练集和验证集的偏倚校正曲线与实际曲线一致性良好，且均接近理想曲线；决策曲线显示训练集阈值概率在0.02 ~ 1.0、验证集阈值概率在0.04 ~ 1.0时，模型均能产生更好的临床效益。 结论 本研究基于体表心电图指标及临床指标构建的房颤伴心力衰竭患者RFCA治疗后复发的预测模型具有良好预测效能，有助于早期评估患者的复发风险。

Objective To develop and validate a nomogram model for predicting recurrence after radiofrequency catheter ablation （RFCA） in patients with atrial fibrillation and heart failure using body surface electrocardiogram indicators and clinical indicators. Methods We retrospectively analyzed 305 patients with atrial fibrillation complicated with heart failure who underwent RFCA from January 2019 to January 2024. Patients were randomized into training set （213 cases） and validation set （92 cases） at a ratio of 7∶3 and followed up for at least 1 year. Based on the recurrence status， the patients were divided into recurrence group and non-recurrence group， with body surface electrocardiogram indicators and clinical indicators collected. Multivariate logistic regression analysis identified for risk factors for post RFCA recurrence， which were used to construct a nomogram. Model performance was assessed using the area under the receiver operating characteristic curve （AUC）， Hosmer-Lemeshow test， calibration curves， and decision curve analysis （DCA）. Results Among the 305 patients， 84 （27.54%） experienced recurrence after treatment. In the training set， 61 patients had recurrence and 152 did not. No statistical differences were observed between the training set and the validation set （all P > 0.05）. In the training set， the recurrence group exhibited a higher proportion of persistent atrial fibrillation and significantly higher CHA2DS2-VASc scores， larger left atrial diameter， longer PR interval， and higher levels of NLR and NT-proBNP compared to the non-recurrence group （all P < 0.05）. Multivariate stepwise regression analysis revealed that high CHA2DS2-VASc score， long left atrial diameter， prolonged PR interval， and high NLR were independent risk factors of recurrence after RFCA （P < 0.05） A four-factor prediction model was established as： Ln （P/1-P） = -12.87 + 0.84*CHA2DS2-VASc score + 0.11* left atrial diameter + 0.03*PR interval + 0.31*NLR. The training and validation models showed AUCs of 0.85（95%CI： 0.80 ~ 0.91） and 0.85 （95%CI： 0.76 ~ 0.94）， respectively， suggesting that the model had good predictive efficiency. Hosmer-Lemeshow test results （χ2 = 2.43， P = 0.965 for the training set； χ2 = 5.30， P = 0.725 for the validation set） confirmed model fit， indicating that the fitted probability value was consistent with the actual probability value. Calibration curves after 1 000 times of Bootstrap repeated sampling showed the bias calibration curves of the training set and the validation set had good consistency with the actual curves， both close to the ideal curve. DCA revealed clinical utility across a wide threshold probability range （0.02 ~ 1.0 for the training set； 0.04 ~ 1.0 for the validation set）. Conclusion This nomogram， based on body surface electrocardiogram indicators and clinical indicators， effectively predicts post-RFCA recurrence in atrial fibrillation and heart failure patients， offering a useful tool for early assessment of recurrence risk.